Reddit User Builds MiniMax H3 Faceswap Workflow in ComfyUI, Expression Sync Is the Hard Part
ShroakzGaming · reddit · 2026-09-05
Reddit user ShroakzGaming shared a faceswap test built on top of MiniMax H3's ref-to-video workflow in ComfyUI, a modified take on the faceswap setups circulating in the community. The main challenge was keeping facial expressions matched to the original footage, which the current version handles well.
The author says the workflow still needs refinement and plans to release the full workflow and tutorial once it's ready, while also testing other faceswap approaches in ComfyUI.
Related event: Users Test MiniMax H3 Video Face Swap at ~90% Accuracy(2 posts)→
More from Multimodal
- Kling AI officially launches Kling 4.0: 15 references per task, 30s clips, 10 keyframes — testingcatalog · 2026-10-23
- Kling 4.0 hands-on: multi-shot cyberpunk robot demo shows strong character consistency — testingcatalog · 2026-10-23
- Run Lola Run director Tom Tykwer uses Seedance in hybrid AI filmmaking experiment — Uncanny_Harry · 2026-10-23
- "Still Worth Something" S01E01: a Pixar-style fully AI-animated comedy short — LPGroves · 2026-10-23
- Music publisher Reservoir sues Suno and Anthropic over AI music training — TinfoilTricorn · 2026-10-23
- Halloween AI short: "Careful who you bring home" — glagolef · 2026-10-23